Describe your equipment. Get a working monitor.
Type one sentence about your machine and its limits. Sentrel's LLM turns it into a typed data schema and live alert rules — no tag database, no scripting, no config files to hand-edit.
The old way: weeks of tag wiring before a single alert fires
Standing up monitoring on legacy gear usually means building a tag database by hand, declaring every data type and unit, then writing scripts or expressions to turn raw values into alerts. It's slow, error-prone, and locked behind a specialist who knows the tooling. The engineer who actually understands the pump waits in a queue while someone else translates intent into config — and a typo in a tag definition can silently break a whole view.
- Manual tag databases: one row per point, types and units declared by hand
- Alert logic written in scripts or vendor expression languages
- Config lives in files only an integrator can safely touch
- Mistakes surface weeks later as missing data or dead alerts
How it works in Sentrel: one sentence in, a typed schema and rules out
You write what you'd say out loud — "Monitor a pump: RPM, pressure, vibration, online status; alert if RPM > 3000 or pressure > 150 PSI." Sentrel's LLM reads that and generates a typed data schema (RPM and pressure as numbers, online status as a boolean) plus the matching alert rules, then shows it to you to confirm. From there the schema validates every incoming reading at ingest and the rules run on the per-reading rule engine, so alerts fire the moment a value crosses a limit.
- Plain English in — equipment, fields, and thresholds in one sentence
- Typed schema out — inferred data types, units, and ranges you can review
- Alert rules generated alongside the schema, wired to the rule engine
- Refine in plain English: add a field or change a limit and regenerate
Why it beats the alternative
Ignition and AVEVA PI assume an integrator and a tag-modeling project; ThingsBoard and a Telegraf+InfluxDB+Grafana stack assume you'll assemble and script it yourself. Sentrel compresses that into minutes and hands it to the person who knows the machine, not a contractor. Because the output is a real typed schema enforced at ingest — not a loose pile of tags — bad data is rejected at the door instead of corrupting your history, and the same description that built the monitor documents it.
- Minutes to a live monitor, not a 6-18 month integration
- Owned by your plant engineer — no integrator or scripting required
- Typed schema enforced at ingest rejects malformed readings
- Your description is the spec, the config, and the documentation
Plain-English Setup, answered.
No. Sentrel's LLM infers types and units from your description — RPM and pressure as numbers, online status as a boolean, vibration with sensible ranges — and shows you the typed schema to review before you accept it.
It becomes the contract for incoming data. Every reading is validated against it at ingest, so malformed values or wrong data types are rejected rather than silently polluting your history and dashboards.
Describe the change in plain English and regenerate — for example, add vibration limits or raise the RPM ceiling. There's no config file to hand-edit and no integrator to schedule, so the engineer who knows the machine owns the change.
More product capabilities
View allStop integrating. Start monitoring.
Describe what you want to watch, connect the gateway you already own, and be live this afternoon — no integrator, no proprietary hardware, no six-month project.